Free Cookies in Training?!
What do you do in a training session when your dog makes a mistake? You gave one cue and your dog does a completely different behavior!
How do you keep the dog in the game?! How do you keep them from spiraling up in frustration?
One answer that is gaining in popularity is giving "free" cookies. Noncontigent reinforcement!
Usually, this is in the form of what is referred to as a "reset cookie." The dog makes an error and the handler tosses a cookie for the dog to chase. This interrupts any behavior the dog did instead and might also serve to set the dog up to try again.
In my classes, I tend to see students on 2 very different sides of this technique.
One group of students are flinging out reset cookies left and right. Sometimes it seems like their dogs get more cookies for a mistake then they do when correct!
And another group is afraid to do any reset cookies for errors. They don't want to risk reinforcing the wrong behavior.
And they're right. A reset cookie DOES reinforce the behavior the dog was doing when you toss the cookie! And I sometimes even see small chains of behavior created where the dog seems to actually offer a specific behavior in an attempt to get that reward. Dogs do what is reinforced!
But despite those risks, I still use reset cookies.
My main focus is on the dog's attitude. Why one can easily argue that you shouldn't need to give free cookies to keep a dog happily working, I find that this is an easy response for us to do with our imperfect training setups.
I'd love to set up a training plan where mistakes don't happen. I appropriately break things down for the dog and things just magically flow. But in reality, I make mistakes in judgment and then mistakes from my dog happen. A quick reset cookie can help to get the dog moving and ready to try again! And it helps to combat the inevitable sigh that I try to hold in to not show my dog my own frustration with myself!
Training plans aren't perfect and they don't have to be!
I don't (usually) see the downsides from these reset cookies that I see some others getting stuck with. The error of the dog repeatedly making the same mistakes and getting reinforced for it. Digging a hole that can be hard to get out of!
Reevaluating Your Training Plan
The first thing to take a look at is your actual training plan. Are you using reset cookies frequently in your session? Several times in a row?
If you find yourself doing "too many" in your short training session then pause. Put your dog on a mat for a second to think, or even better end the training session or switch to a new behavior entirely so you have more time to think about what to do in your next session. You do not have to end on a good note!!!
Rewarding a few errors is not going to make that big of a difference for what you're trying to teach. But repeatedly rewarding different things is not only making it more likely to occur again, but the dog is also likely getting confused with the lack of clear criteria. Sometimes doing X behavior leads to a thrown treat and sometimes it doesn't! That's confusing and frustrating! You want your reset cookies to reduce frustration from building, not create more of it!
Cuing Other Behaviors?
What about cuing another behavior to reset the dog and then rewarding that?
Sure! The most common example is cuing the dog to do a hand touch and then giving either a treat from your hand or a cookie toss to reward the touch. You will see me do this option in my training too!
Yet you have to be aware that while the problematic behavior is now one step removed from the actual cookie, you still have a mini behavior chain. Assuming that you taught your hand touch (or other behavior) with positive reinforcement, the cue to touch your hand will reinforce the dog's error! Even if you don't follow up that hand touch with a treat, the dog is still getting reinforcement for the error. That doesn't mean you shouldn't do it! It does mean that looking at how you're splitting the behavior you're trying to teach is the most important aspect!
Placement of Reset Cookies
When you do decide to toss a reset cookie, look at where you are tossing it. Is there a way you can toss the treat to help your training plan? Or at least not make the error more likely to happen?!
One example that comes to my mind is pivoting. If I'm working on the dog's butt coming into heel position, then I'm more likely to toss a reset cookie so that the dog has to do a turn to their left in order to get it. Even better if they have to do a 180 turn to the left. I deliberately move my hand to help that left turn over the right turn to get that treat behind them as this left spin mimics some of the action of that counter clockwise pivot towards me!
A recent example I've seen is with front training. The dog had a bit of a space bubble and of course also loved thrown cookies. I got the impression that the dog was stepping off the target in front of the handler in expectation of a cookie tossed behind him. Along with some other advice on trying to break things down a bit more to get that duration in front, I suggested resetting by having the dog go through the handler's legs. Either a cookie tossed straight behind through the legs, or having the dog go through and wrap around a leg in order to get a cookie that would then be tossed back out in front. This should make it less likely the dog backs off the target or even weight shifts backward in expectation of where the cookie might appear!